Intraseasonal Variability of Summer Monsoon Rainfall and Droughts over Central India
Rainfall over Madhya Pradesh (MP) in central India has large intra-seasonal variability causing droughts and floods in many years. In this study, rainfall variability in daily and monthly scale over central India has been examined using observed data. Consistency among various datasets such as rainfall, surface temperature, soil moisture and evapotranspiration has been examined. These parameters are from various different sources and critical for drought monitoring and prediction. It is found that during weak phases of monsoon, central India receives deficit rainfall with weaker monsoon circulation. This phase is characterized by an anticyclonic circulation at 850 hPa centered on MP. The EOF analysis of daily rainfall suggests that the two leading modes explain about 23–24% of rainfall variability in intraseasonal timescale. These two modes represent drought/flood conditions over MP. Relationship of weak phases of rainfall over central India with real-time multivariate (RMM) indices of Madden Julian Oscillation (MJO) has been examined. It is found that RMM-6, RMM-7, RMM-1 and RMM-2 describe the weak monsoon conditions over central India. However, frequency of drought occurrence over MP is more during RMM-7 phase. Surface temperature increases by about 0.5°–1° during weak phases of rainfall over this region. Soil moisture and evapotranspiration gradually reduce when rainfall reduces over the study region. Soil moisture and evapotranspiration anomalies have positive pattern during good rainfall events over central India and gradually reduce and become negative anomalies during weak phases.
KeywordsRainfall Central India drought intraseasonal interannual MJO
Droughts cause significant loss to economy and damage to environment. Droughts cannot be prevented, but much can be done to reduce the impacts through preparedness, mitigation using better forecasting techniques. Droughts occur due to deficient precipitation and high evapotranspiration. Central part of India, also known as the core monsoon zone (CMZ), receives most of precipitation during summer monsoon season [June, July, August and September (JJAS)]. The CMZ region is considered representative for both the mean performance as well as for variability of the monsoon over India (Sinha et al. 2011). The population of the Indian subcontinent depends on cereal and pulse production. Central India is counted as one of the biggest suppliers of wheat, soybean, paddy, etc. The summer monsoon rainfall is also important for fulfilling the demands of drinking water and agriculture. Drought disasters often cause reduction in crop production or even maximum crop failure over central India. This region receives almost 95% rainfall during southwest monsoon (JJAS). However, this region is vulnerable to extreme events such as droughts and floods and experiences regular, spatially broad, and long-term droughts that cause the most severe losses to the agricultural economy (Sikka and Gadgil 1980; Rajeevan 2001; Shrivastava et al. 2016a).
Rainfall variability in monsoon months over India is the result of large-scale monsoon variability in seasonal to inter-annual timescales as well as intraseasonal oscillation. The Asian monsoon is affected by various factors like El Niño-Southern Oscillation (ENSO), North Atlantic Oscillation (NAO) and the Indian Ocean Dipole (IOD). Impact of ENSO on the Indian summer monsoon rainfall has been studied by several researchers since long (Sikka and Gadgil 1980; Shukla 1987; Kar et al. 2001; Kar 2007). The IOD and the ENSO affect the Indian monsoon at interannual time scale (Ashok et al. 2001). The predictions of the Indian summer monsoon rainfall (ISMR) and role of SSTs during monsoon 2009 have examined by Acharya et al. (2011). Active (wet) and weak (dry) phases of rainfall have been examined in various studies such as Goswami and Mohan (2001), Gadgil and Joseph (2003), Goswami and Xavier (2003), Pabón and Dorado (2007) and Shrivastava et al. (2016b) and many others. The intraseasonal variability of rainfall is associated with northward propagating large-scale intraseasonal oscillation (ISO) with a 30- to 60-day period, which also controls the onset of southwest monsoon season over India (Yasunari 1979; Lau and Chan 1986; Kar et al. 1997; Gadgil 2003). These are also associated with eastward propagating Madden Julian Oscillations (MJO) as suggested by Kar et al. (1997). Using daily rainfall data and real-time multivariate MJO indices (RMM) of Wheeler and Weickmann (2001), the intra-seasonal variation of daily rainfall distribution over the Indian region associated with various phases of eastward propagating MJO was examined to understand the mechanism associating the MJO to the intraseasonal variability of the Indian monsoon (Pai et al. 2011).
An increase in temperature affects crops and crop yield. Previous studies (Sun et al. 2005; Lucas-Picher et al. 2011) have found more surface temperature over the Indian region during drought conditions. Soil moisture information is important to understand drought and other extreme events because soil moisture and rainfall have one to one relationship. Many researchers have used soil moisture data in their study to understand the climate variability (Sun et al. 2005; Miralles et al. 2012; Shrivastava et al. 2016b).
There have been several studies on monsoon variability over the monsoon core zone. However, no studies exist on the rainfall variability over Madhya Pradesh to understand the drought occurrence and its mechanism using rainfall, temperature, soil moisture and evapotranspiration. Therefore, in this study, rainfall and its relationship with evapotranspiration, soil moisture, temperature and winds have been studied using various methods over Madhya Pradesh in central India in intraseasonal timescale. In the Sect. 2, various datasets used in this study have been discussed. In Sect. 3, interannual variability, intraseasonal variability, atmospheric circulation anomalies, regression with soil moisture and evapotranspiration are discussed. The last section, Sect. 4 summarizes the main results.
2 Data and Methodology
High-resolution observational and remote sensing datasets are available for precipitation, temperature, soil moisture and evapotranspiration. These ground and satellite observations help to understand ground conditions of these variables. The India Meteorological Department (IMD) is responsible for collection of observational precipitation and temperature in India. IMD-derived gridded datasets provide ground condition of precipitation and temperature and this high resolution gridded datasets developed by Pai et al. (2011) for precipitation and Shrivastava et al. (2008) for temperature. These data have been used in the present study.
Remotely sensed soil moisture data from European Space Agency (ESA) Climate Change Initiative (CCI) and evapotranspiration from moderate-resolution imaging spectroradiometer (MODIS) MOD16 global evapotranspiration (ET) product have been used in this study as a proxy of observation. ESA-CCI has made available soil moisture data from various sensors, i.e., SMMR, SSM/I, TMI, AMI-WS, ASCAT, AMSR-E, WindSat, AMSR2, etc. The data sets have been developed following procedure described in Hain et al. (2011), Parinussa et al. (2012), Liu et al. (2011, 2012), and Wagner et al. (2012) available at http://www.esasoilmoisture-cci.org/. The merged products present volumetric soil moisture (m3/m3) at a spatial resolution of 0.25° by 0.25°. MODIS MOD 16 global evapotranspiration datasets are regular 1-km2 land surface ET datasets at 8-day interval have been used in the present study.
Atmospheric wind data from the European Centre for Medium Range Weather Forecasts (ECMWF) Reanalysis-Interim (ERA-Interim, Dee et al. 2011) have been used for the same period. Global reanalysis data sets (such as the ERA-Interim) provide a long-term and consistent data sets useful for climate analysis and diagnostics. The ERA-Interim reanalysis uses a four-dimensional variational assimilation method and a high-resolution global model at T255 resolution. In this study, reanalysis winds data at 0.25° by 0.25° have been used. The real-time multivariate MJO indices (RMM1 and RMM2) of Wheeler and Hendon (2004) were used for defining the various phases of MJO. The data were obtained from http://www.bom.gov.au/bmrc/clfor/cfstaff/matw/maproom/RMM/.
3 Results and Discussion
3.1 Climatological Feature and Interannual Variability of Rainfall
Correlation of rainfall in June, July, August and JJA season over Central India with that of other regions in India. The regions are identified in Fig. 1
(B) West India
(68°E–74°E and 21°N–30°N)
(C) South Upper India
(72°E–82°E and 16°N–21°N)
(D) South Lower India
(73°E–81°E and 8°N–16°N)
(E) North Lower India
(74°E–82°E and 26°N–30°N)
(F) North Upper India
(73°E–81°E and 30°N–37°N)
(G) East India
(82°E–89°E and 17°N–28°N)
(H) North East India
(89°E–98°E and 22°N–30°N)
3.2 Monthly Composite Analysis
Excess and deficit rainfall years for Central India
1987, 1992, 1996, 2009, 2010, 2012, 2014
1990, 1994, 2001, 2008, 2011, 2013
1984, 1987, 1989, 2002, 2004, 2008
1986, 1994, 1997, 2001, 2005, 2013
1993, 1998, 1999, 2000, 2001, 2005, 2007, 2008, 2009, 2014
1982, 1984, 1994, 2004, 2006, 2012, 2013
Weak monsoon phases over central India
Observed seasonal mean rainfall
28 July–01 August
3.3 Intra-seasonal Variability of Rainfall over MP
3.3.1 Composite Analysis of Weak Phases
Daily climatology of rainfall from June to August has been computed by averaging daily rainfall from 1982 to 2014 from IMD-derived gridded data (at 0.25° × 0.25°). Daily anomalies of rainfall have been calculated by difference between daily rainfall and daily climatological mean. Table 3 shows weak phases of rainfall (JJA) from 2002 to 2011 over the study region. Negative rainfall anomalies for five or more days have been counted as a weak phase of rainfall. It is found that 2004 is normal rainfall year but many weak phases of rainfall have been observed in this year. The year 2008 is normal season of rainfall but two long weak phases from 14 to 24 July and from 18 to 25 August are noticed. Therefore, one or two extreme rainfall events affect the seasonal mean rainfall and even if there are long weak phases, seasonal mean rainfall may become normal. Table 3 represents the seasonal mean of rainfall. 2002, 2007, 2009 and 2010 were deficit and 2006 and 2011 were excess years over central India. In this study, many long weak phases were found over the study region. Following the same method as for rainfall, daily anomalies of zonal and meridional components of winds at 850 hPa have been computed and analyzed as described below.
In 2002, two long weak phases with weaker south westerly wind anomalies were seen in July over MP. In the first weak phase, an anti-cyclonic circulation anomaly at 850 hPa over south India is noticed with 10–15 mm day−1 deficit of rainfall during 02–11 July. In the second weak phase, an anti-cyclonic anomaly at 850 hPa over south MP is noticed with 10–15 mm day−1 deficit of rainfall during 21–31 July. In both the cases, one cyclonic anomaly over Pakistan region is seen. The year 2003 (a normal rainfall year) has one weak phase over central India in August. Strong cyclonic anomaly near Andhra coast over the Bay of Bengal is noticed with weaker south westerly winds over central India. This weak phase brings 6–10 mm day−1 deficiency in rainfall. In 2004, two weak phases in June and July were identified over central India. Strong westerly and weaker south westerly at 850 hPa were seen during 21–30 June and 10–16 July. An anti-cyclonic anomaly over south of MP with 6–8 mm day−1 rainfall deficit during 21–30 June and an anti-cyclonic anomaly with 6–10 mm day−1 rainfall deficit over central MP is also seen during 10–16 July. In 2005, two weak phases were noticed in August. In the first weak phase, an anti-cyclone anomaly over central India and strong westerly anomaly over north India with 10–15 mm day−1 is seen during 07–15 August. In the second weak phase, an strong easterly anomaly over south India, north westerly anomaly over north India, and an anti-cyclonic anomaly at 850 hPa with negative rainfall anomaly of 8–10 mm day−1 over central India has been noticed during 23–31 August. The year 2006 was an excess rainfall year and no weak phase was identified. In 2007, two weak phases during July and August have been seen. In the first weak phase, strong easterly anomaly over south India, strong north westerly anomaly over north India and anti-cyclonic anomaly at 850 hPa with 15–20 mm day−1 deficiency of rainfall over central India is seen during 18–23 July. In the second weak phase, strong south westerly with anti-cyclone over Western Ghats and the Arabian Sea at 850 hPa with 10–15 mm day−1 deficiency of rainfall over central India has seen during 10–20 August. In 2008, two weak phases in July and August were seen. In the first weak phase, strong anti-cyclonic anomaly covers the east and central India and cyclonic anomaly near to equator (5°N) at 850 hPa with 10–15 mm day−1 deficiency in rainfall over central India has been seen during 14–24 July. In the second weak phase, a ridge over central India at 850 hPa with 8–10 mm day−1 deficiency in rainfall over central India is seen during 18–25 August. The year 2009 was a deficit year but only one long weak phase has been noticed. In this weak phase, an anti-cyclonic anomaly over central India, strong north westerly with 15–20 mm day−1 deficit rainfall has been noticed over central India.
3.3.2 Empirical Orthogonal Function (EOF) Analysis
3.3.3 Rainfall Anomalies and MJO Indices
As seen in the previous section, monsoon rainfall over central India exhibits strong intraseasonal variability causing droughts or floods. Madden Julian Oscillation (MJO) is one of the most influencing factors of the intraseasonal variability of the monsoon rainfall over India. Pai et al. (2011) using IMD daily gridded rainfall data and Wheeler–Hendon MJO indices, examined the intra-seasonal variation of daily rainfall distribution over India associated with various phases of eastward propagating MJO life cycle. They have found that during MJO phases of 1 and 2, formation of positive convective anomaly over the equatorial Indian cause break monsoon type rainfall distribution over India. As the MJO propagates eastwards to west equatorial Pacific through the maritime continent, a gradual northward shift of the convective activity over the Indian Ocean is observed. During phase 4, the northward propagating convective zone merges with monsoon trough and enhances rainfall activity over the region. During phases 5 and 6, the patterns are reversed compared to that during phases 1 and 2 and India experiences active monsoon conditions. During the subsequent phases (7 and 8), the convective anomaly patterns are very similar to that during phases 1 and 2. A general decrease in the rainfall is also observed over most parts of the country. However, it is not clear from the above study and similar past studies if the real-time MJO indices represent drought conditions over MP and if so, which phases of MJO need to be monitored or forecasted.
3.3.4 Soil Moisture and Evapotranspiration
Rainfall variability (in daily and monthly scale) in the context of drought occurrence over Madhya Pradesh in central India during summer monsoon has been studied using observed datasets. Mechanism of such rainfall variability has been examined so that an effective drought monitoring system can be developed for the study region. Consistency among various datasets such as rainfall, surface temperature, soil moisture and evapotranspiration have been examined. These parameters are from various different sources and critical for drought monitoring and prediction. Main conclusions of the present study are the following.
Composite analysis of winds 850 and 200 hPa during excess and deficient rainfall years show that the southwesterly winds are stronger and a cyclonic circulation is formed at 850 hPa over central India during excess rainfall months. The Tibetan anticyclone is shifted west and northwards from its normal position when the study region gets above normal rainfall. The opposite happens when the regions get below normal (deficit) rainfall. During weak phases of monsoon, central India receives deficit rainfall with weaker monsoon circulation. This phase is characterized by an anticyclonic circulation at 850 hPa centered on MP. The EOF analysis suggests that the two leading modes explain about 23–24% of rainfall variability in intraseasonal timescale. These two modes also represent large-scale monsoon variability with central India and Western Ghat regions having same phase and the north-east India having opposite phase. These two modes also represent drought conditions over MP.
Relationship of weak phases of rainfall over central India with real-time multivariate (RMM) indices of MJO has been examined. It is found that RMM-6, RMM-7, RMM-1 and RMM-2 describe the drought conditions over central India. However, frequency of drought occurrence over MP is more during RMM-7 phase. Therefore, a relationship between droughts over MP and the eastward propagating MJO mode has been established. Surface temperature increases by about 0.5°–1° during weak phases of rainfall over this region.
An attempt has been made to document relationship between remotely sensed soil moisture and evapotranspiration with meteorological conditions over central India. It is found that the soil moisture and evapotranspiration gradually reduce when rainfall reduces. Soil moisture and evapotranspiration anomalies have positive pattern during good rainfall events over central India and gradually reduce and become negative anomalies during weak phases.